XP or not : no utter software development methodology

I've practiced XP for about four years, and I believe it's a excellent methodology for developing robust software, but recently I doubt what I got from XP, it become even worse when I read "Code Complete 2nd Edtion"(CC2E) those days

- XP said unit testing is very very very important, so we should testing first, CC2E said test is important but not so much

- XP said write code immediately, and improve software during iteration, but CC2E said if you don't get a clear/detail requirements before starting, you will be a loser, and improve software during iteration isn't a good way to follow, if we can do it well at starting, then it's the best way

- From my practice, I met delay delivery several times even we use XP, and also a bad stable status of our software, I know it's not all fault of XP, but from CC2E I do know XP isn't a silver bullet too, XP isn't suitable for every type of software development, it has its own suitable area

其实你已经非常接近答案了。Steve McConnell是微软员工，而微软是一个做产品的公司。而XP是从C3项目脱胎出来的。这就已经很明显了。

Do you mean XP isn't suitable for product development but some framework/library/tools etc.?

From a table of CC2E, it said XP suitable for internet/intranet, etc. but not suitable for boxed software(product), if so, that's a great pity!

项目。XP完美适合的是项目开发，尤其是内部IT项目，ERP/CRM之类的。

对于软件产品，它不是不适合——它的很多最佳实践例如测试驱动、持续集成、结对编程等等仍然有效。它只是不足够。你还需要别的方法学和最佳实践。稍后我会写一个blog来详细说这个事情。

I do know you are an employee of ThoughtWorks(A XP company?), some of your softwares like "CruiseControl", "Selenium" are product or project? or how do you treat them inside your company?

I wish to read your blog...

我们的实践证明，XP很适合软件产品。
当然了，XP不是银弹，不能解决delay delivery。

话说回来， 微软的产品也是经常延迟的，
SO，没有任何证据表明微软的方式比XP好。

that's fine, could you please share some of your experiences here?

写好了
http://gigix.thoughtworkers.org/2007/11/13/improving-agile-in-product-designing

所以，最终不是要不要敏捷的问题，而是能不能在敏捷的成绩单上睡大觉的问题——显然答案是不能